We’ve moved!

Thanks to the work of John Quillen, we have implemented a site dedicated to the Sandy Hook Project.  From here on out I will be posting all of my articles and analysis regarding gun violence at http://www.thesandyhookproject.org.  Over time, we will also be publishing more links to research on this site as well.

Posted in Sandy Hook Project | Comments Off

Struggling with the Tiahrt Amendment

Previously I had mentioned a NY Times article about the NRA’s efforts to suppress data they consider harmful to the support of gun rights.  I recently experienced this challenge first hand when trying to understand the movement of guns from state to state.

While researching the impact of the Brady Handgun Violence Prevention Act, I found an analysis performed in 2003 by Dr. Phillip Cook.  In this analysis, Dr. Cook compared the levels of violence in states that required background checks before the Brady Act went into effect to those that had only instituted background checks as a result of the Brady Act.  Based on this comparison he concluded that the Brady Act has had no measurable impact on gun-related homicides.

Because of the relative ease with which people can move between states, I have wondered whether state by state comparisons are completely valid.  I decided to explore the movement of guns between states to determine if it is significant enough to affect such comparisons.

I began extracting gun tracing data from the ATF web site.  The tracing program tracks the movement of firearms and identifies the state of origin of firearms recovered from arrests.  One of the challenges I encountered is that the data on the website is highly summarized and only provided in PDF documents.  I found this experience very different from other Department of Justice websites that provide direct access to more granular data.

As I wrestled with extracting data from the PDF’s, I discovered that these impediments were actually mandated by law.  They are the result of the Tiahart Amendment passed in 2003.  This amendment limits access to detailed tracing data to registered law enforcement agencies only.  According to an NRA sponsored web site, the Tiahrt Amendment prevents the the use of this data by gun control advocates “…in questionable studies used to support their goals“.  While the NRA cites the support of the ATF and the Fraternal Order of Police for limiting access to this data, there are also law enforcement and government organizations opposed to the amendment, including The Mayors Against Illegal Guns.

I find this control of data extremely disturbing.  I understand that there might be limitations about what can be inferred from the data, but I don’t believe that this is sufficient justification for preventing access to that data.  Scientists and statisticians are always required to account for such possibilities in their research.   I am deeply concerned that in pursuit of protecting second amendment rights, there are those who are willing to subvert the individual’s right to make decisions based on the most accurate and complete data available.  Which is more dangerous to this country: gun control or information control?


Posted in Uncategorized | Comments Off

Exploring the Declining Trend in Homicides

One of the most surprising discoveries from my research on gun-related violence has been the decline in US homicide rates over the past twenty years.  My general impression had been the opposite, especially given the series of mass shootings that occurred in 2012.  However when we look at the trend for the past fifty years, we observe that homicide rates in 2010 are actually lower than homicide rates experienced during the 1960′s.   We can see this appears to be a generally consistent experience across the US when we look at not only the US homicide rate, but also at those states that experienced the highest homicide rates during this period.

There appear to be three distinct periods during which homicide rates declined: 1974-1976, 1980-1985 and 1991-2010.   This most recent period of decline follows a spike in violence during the late 1980′s that law enforcement agencies described as the crack cocaine epidemic.  During this period there were increases in most forms of crime in the major US cities.  This trend reached its peak in the US in 1991, though some cities did not peak until 1993.  Starting in 1992, the US homicide rate began to decline, dropping in thirteen out of the following nineteen years.  On average, the US homicide rate declined 3.6% per year during this period.

A number of  theories exist regarding the causes for this reduction, from law enforcement initiatives to a reduction in lead poisoning.  In 2000, former President Bill Clinton attributed reductions in gun-related crime to the enactment of the Brady Bill which he had signed into law in 1993.

Is this claim credible?  In prior postings I explored the impact on homicide rates of legislation both limiting and enabling citizens to publicly bear arms.  Neither approach appeared to have a significant effect on homicide rates in their specific locales.  But the Brady Bill is not about blanket bans on firearms.  Is it possible that measures such as mandatory background checks and greater controls on gun dealers have had a greater impact on gun-related crime? Or is the reduction in homicides a result of factors independent of the guns themselves?  In my next posting I will seek to determine if any relationship can be established between the reduction in homicides and the introduction of this law.

The data employed in this analysis can be found here.

Posted in Sandy Hook Project | Tagged , , | Comments Off

Concerns About “Death by Hammer” Might be Misplaced

The Newtown massacre drew immediate public demand for new gun control and bans on assault rifles.  In response to this outcry, many gun rights advocates sought to change the conversation, focusing on underlying causes of violence rather than the guns themselves.  In some cases, these advocates put forward intelligent and factual arguments.  In others, however, efforts to down play relationships between guns and violence involved highly misleading information.  For example, here is an info-graph that has emerged on the Internet.  This same message is the subject of this Fox News article.

The data presented here seems to indicate that hammers cause more deaths than assault weapons.  One might also extrapolate from this piece that gun-related violence is a relatively small issue, on scale with other types of violence, such as bludgeoning or stabbing.

The actual source of this data is not provided, but these numbers do correspond with data provided to me by my colleague, Dr. Ken Collier.  Ken pointed me to a table on the U.S. Department of Justice / Bureau of Justice Statistics website.  This table contains homicide aggregates for 2011, categorized by motive and method.  Assuming that the individual who created this graphic was using this table, the numbers presented correspond to the total number of people murdered in 2011 by rifles and the total number of people murdered in 2011 by blunt instruments (which includes hammers).  What is important to note is what has been overlooked in this graphic.  The number of homicides committed by rifles is only a small part of the picture.

Firearm type Homicides
Rifles 323
Shotguns 356
Handguns 6,220
Other guns/type not stated 1,684
Total gun-related homicides 8,583

Those 323 homicides represent only four percent of all firearm-related homicides in 2011.  If we take all firearm-related homicides into account, we derive a very different set of ratios between firearm-related deaths and “death by hammer”:

  • Of the 12,664 recorded homicides that occurred in 2011, 68% of those were caused by firearms.  More than two out of every three homicides was committed using a firearm.  This is a ratio that has remained relatively constant for the past 30 years.
  • In comparison, only 13% of homicides involved a sharp instrument; knives, scissors, box cutter, etc.  Homicide victims in 2011 were five times more likely to be shot to death than stabbed to death.
  • Only 4% of homicides in 2011 involved blunt instruments such as hammers, clubs, etc. Homicide victims in 2011 were seventeen times more likely to be shot to death than bludgeoned to death.

There is potential validity to the argument that assault rifles represent a tiny fraction of homicides committed in the U.S., given that a large percentage of homicides (72%) are perpetrated with handguns.  However, the important point that we must continue to recognize, and that the author of the Fox News article fails to note, is that gun-related violence is a major issue in this country and the relationship between firearms and violent crime must be explored and addressed.

Posted in Sandy Hook Project | Tagged , , | Comments Off

Impact of Gun Control Legislation in Washington DC

In my last posting I had demonstrated that “Right to Carry” legislation had no verifiable impact on the homicide rate in those states that had adopted it.  A number of people who read that posting asked if I would perform a similar analysis regarding the impact of gun control legislation.  If enabling citizens to bear arms in public had no demonstrable impact on violence, would removing their ability to bear arms have any impact?  And if gun control did have an impact, would it be positive or negative?Two cities in the US (Chicago and Washington DC) have enacted legislation that specifically bans the ownership of handguns.  In both cases, those laws were eventually found unconstitutional and overturned.   I will first focus on Washington DC.  In 1975, The Firearms Control Regulations Act was established.  This law prohibited residents from owning handguns, automatic weapons and high-capacity semi-automatic weapons.  Those citizens who had purchased and registerted handguns prior to 1976 were allowed to keep them but were not allowed to carry them in public.  This law was challenged on the basis that it violated citizens’ second amendment rights and was overturned in 2008.The goal of this analysis is to determine if any conclusions can be drawn from the roughly thirty years that this law was in effect.  I will explore two scenarios:

  • Did preventing citizens from bearing handguns reduce the levels violence in Washington DC?
  • Conversely, did removing criminals’ concerns about citizens being able to defend themselves actually lead to an increase in violence?

As in my last analysis, I have selected the overall homicide rate (homicides per 100,000 people) as the figure of merit.  While assessing the specific impact on firearm-related homicides and violence would be enable us to draw more specific conclusions, the intent of these laws is to have an overall impact on violence.  If gun control legislation only impacts the method by which people killed, then such legislation is ultimately ineffective.The statistics for homicides in Washington DC have been extracted from the US Justice Department’s Uniform Crime Reporting Statistics site.  The following graph depicts the homicide rate for the District of Columbia from 1965 – 2010.

It is immediately evident that Washington DC experienced a massive spike in homicides during this 30-year period.  Based on this data, it would be impossible to make any claims that the introduction of the Firearms Control Regulations Act had a positive impact on violence.   The question is, does this data indicate that the removal of a citizen’s right to bear arms lead to an uptick in violence?I would argue it does not.  While there is clearly a dramatic jump in homicide rates from 1986 to 1991, that trend begins to reverse itself in 1992 and is essentially negated by 2005.   If we compare the homicide rate for the year prior to the law going into effect (1975) and the first full year after the law was overturned (2008), there was a net decrease in homicide rates of 4%.  Since a simple year over year comparison can be misleading, I calculated the mean homicide rate for the years 1960 to 1976 and the mean homicide rate for 2008 to 2010. The mean homicide rate for the years preceding the gun ban was 24.3.  The average homicide rate for the years following the law’s repeal was 25.9.  This represents a 6.57% increase.If we take the set of the homicide rates for the years preceding and following the ban, we see that the mean homicide rate was 24.7, with a coefficient of variance of 36%.  Given the small population size and the degree of variance, a 7% fluctuation cannot be considered statistically relevant and it is impossible to derive any conclusions from the change in homicide rates.

Because the trend was corrected well before the reinstatement of 2nd amendment rights, there is no real correlation between the ability to bear arms in public and the increase in crime.  What did cause this bump?  The jump experienced during this period has been attributed to illegal drug use, particularly the “crack epidemic” of the 1980’s through 1990’s.   Increases in major crime rates (not only homicides) during this period were particularly prevalent in areas of higher poverty, such as Washington DC.  I have started to validate this hypothesis by correlating homicides and drug-related crime, but obtaining a quantitative measure of the extent of drug crime is proving challenging.Even without establishing this relationship, I think it is safe to say that Washington DC’s efforts to band handguns had no demonstrable impact on the homicide rate in that city.   This seems reasonable, as the law did little to eliminate the existing supply of handguns from the criminal population, or even reduce the number of legal gun owners as of 1976.  Furthermore, gun control legislation at the municipal level does virtually nothing to prevent individuals from procuring guns in nearby areas without gun control legislation.  Until gun control legislation is ubiquitous and includes measures to reduce the number of illegal guns in circulation, it will probably be very challenging to demonstrate whether such legislation is effective.At the same time, this data refutes the argument that an armed public is required to deter crime. While the crime rate significantly increased shortly after the ban went in place, an equally significant reduction was also achieved while the ban was in place.  If armed citizens are critical to deterring crime, then rates should have continued to increase until the law was repealed.

Posted in Sandy Hook Project | Tagged , , | 1 Comment

Constraints on Research Regarding Gun Ownership and Violence

Cathy O’Neil, a colleague and collaborator on the Sandy Hook Project, pointed me to this article in the New York Times.  I think this powerful quote highlights the primary reason I started work on the Sandy Hook Project:“In the absence of reliable data and data-driven policy recommendations, talk about guns inevitably lurches into the unknown, allowing abstractions, propaganda and ideology to fill the void and thwart change.”What limited data and analysis that exists is often misrepresented and abused, and is given much less attention than anecdotal evidence.  It is relatively simple to produce a handful of cases that support either side in this debate.  What we really need is to understand the true impact of guns on our society.  Push back by the NRA that any such research would be “political opinion masquerading as medical science.” is unacceptable.  We can only make intelligent decisions when we have the fullest picture possible.

Posted in Sandy Hook Project | Tagged , | 1 Comment

Investigating Statistics Regarding Right to Carry Laws

“The only thing that stops a bad guy with a gun is a good guy with a gun”

Wayne LaPierre, EVP, NRA

A point made by many gun rights proponents is that firearms are essential for self-defense.  Many have argued that removing firearms from the public would result in higher crime rates, as criminals would be less concerned about potentially lethal consequences.  To demonstrate this, gun rights advocates will point to the introduction of “right to carry” legislation; laws that defend the rights of gun owners to carry and conceal weapons on their person in public.  Supporters of right to carry legislation will point to crime statistics from states that have either banned concealed weapons or to crime statistics from those states that have instituted right to carry legislation to support this argument.After reading one such claim, I started researching these statistics online.  The first site I was directed to was justfacts.com.  Justfacts.com is a site that provides data and some analyses on a range of social topics.  While the mission statement of the site is to present unbiased reporting of facts, the authors of the site openly admit to conservative perspectives.  The authors indicate that they intentionally do not provide analyses or assessments of their findings so as to not persuade readers in a specific direction.  They prefer that the facts speak for themselves.There are three graphs on the JustFacts site that I selected for analysis.  These graphs depict the homicide rates (homicides per 100,000 people) of three states: Michigan, Florida and Texas.  I have pulled the following images from their site:Note that in all three cases, the authors pointed out that each state’s homicide rate was, on average, some percentage lower after the right to carry law was put in effect.  Furthermore, that reduction was greater than the average drop experienced in the US overall.  These statements would seem to imply that the introduction of right to carry laws has had a positive impact on the homicide rate in three states.  Note that the authors do not explicitly make a conclusion, they only point out this difference in each case.I decided to test if there is any validity to such a conclusion.  Was the reduction in homicide rates experienced in these states statistically meaningful? Could a causal relationship be implied (not necessarily proven, just implied) from the changes that occurred after the right to carry laws went into effect?Before continuing I must point out that justfacts.com is using overall homicide rates and not specifically firearm-related homicides.  Optimally justfacts.com would only employ firearm-related homicide rates for this analysis.  Changes in the rate of other methods of homicide such as blunt force trauma could skew the overall homicide rate.  However, I have observed that the ratio of firearm-related homicides to total homicides in the US has remained consistently at 65% over the past 30 years, only deviating significantly due to the events of 2001. Given this, and that my goal is to assess the validity of the findings given the data employed by justfacts.com, I continued using total homicide statistics.Looking at the three graphs, I immediately noticed the following:

  1. All three states appear to have experienced higher homicide rates than the overall US homicide rates until the early to middle part of the 1990’s
  2. By 2000, these three states were much more closely aligned to the US homicide rate

My objective was to measure this change in relation to the dates that Right to Carry laws were enacted in each state.I started by seeking to duplicate the work done by justfacts.com and producing the graphs depicting the individual state homicide rate per 100,000 residents against the US homicide rate over the past 50 years.  The justfact.com authors thoroughly site their data sources.  In this case, the data was provided by the US Justice Department’s Uniform Crime Reporting Statistics site.  Using this site I was able to extract violent crime statistics for each state from 1960 to 2010.  I generated this comparison for each state and committed graphs with these comparisons to the Sandy Hook project repository.Having produced the same results as justfacts.com, I then sought to compare the trends experienced in Florida, Michigan and Texas to the remaining US states.  From the US homicide rate, I could see that there had been a peak in homicide rates in the early 1990’s followed by a steady decline to 2010.  I wanted to understand how consistent this pattern was across states.  This would allow me to assess if the rates experienced in these three states were anomalous or not.

This graph depicts the overall US rate (a weighted average), the mean homicide rate for the set of states (an un-weighted average) and the standard deviation from that mean for each year.I then calculated the difference between the homicide rates experienced each of the three states and the mean over the fifty year period.  This would provide the following information:

  • How the fluctuations in the homicide rate deviate from the rest of the country.  The difference from the mean should identify the occurrence of trends that are potentially significant.
  • Whether the differences displayed are statistically significant.  I have included the standard deviation to provide a frame of reference.  Over the course of the fifty years plotted, one standard deviation contains between 54% and 70% of the states in a given year.  Two standard deviations contain between 94% and 100% of the states in a given year.  This provides a good reference for what is statistically relevant behavior and what is a natural variation.

I started with Texas.  This first graph is similar to the graph published on justfacts.com  This graph demonstrates that until 1996, the homicide rate in Texas was substantially higher than the national average.  This is the same year that Texas enacted the right to carry law.

However, when we look at how Texas differs from the mean homicide rate, we get a different picture:

We can see that the real downward trend preceded the implementation of the right to carry law.  Once the law was in effect, the homicide rate for Texas remained relatively consistent with the mean homicide rate.  It even trends very slightly higher.  The point made by justfacts.com that the “…Texas murder rate has averaged 30% lower than it was before the law took effect, while the U.S. murder rate has averaged 28% lower” is due to the fact that the homicide rate in Texas was considerably higher to begin with.  During the period of 1976 through 1991, Texas was in the top 10% of homicide rates in the nation.  There is no statistical evidence that indicates the right to carry law has had a measurable impact on homicide rates in Texas.The findings for Florida were similar.

Florida experienced a general downward trend starting in 1980, seven years prior to the introduction of the law, but still remains above the mean homicide rate and consistent with the national average homicide rate.

Michigan enacted the right to carry law in 2001.  That was five years after the homicide rate experienced a consistent eight-year downward trend.  Since the enactment of law, there have been some isolated dips but the homicide rate has remained statistically consistent.

Based on this analysis, I see no evidence indicating that the right to carry law has had a material impact on these three states.  As of 2010, these states all remain at or above the national average and above the mean homicide rate among the fifty states.  While their respective homicide rates have improved, in every case these improvements preceded the introduction of the right to carry law.  While the data provided by JustFacts.org is  accurate and the statistics they have generated are correct, their analysis is lacking and fails to paint a full picture.  Their statements could easily be misconstrued by those who do not thoroughly investigate the data behind the statistics.The data for this analysis can be found here.

Posted in Sandy Hook Project | Tagged , , | 1 Comment

A Brief Analysis of Firearm-related Homicide Rates

My first analysis for the Sandy Hook Project involved comparing US homicide rates, and firearm-related homicide rates, with those of other countries.  There were two questions I wanted to explore:

  1. How do the level of gun-related deaths in the US compare to countries of similar levels of prosperity?
  2. Which countries experience similar levels of gun-related deaths as the US?

While these numbers have been presented numerous times in the past, I wanted to ensure that I fully understood the data sources and methodologies employed for calculating the homicide rates.I started with the mortality statistics published each year in the National Vital Statistics Report of Deaths published by the Center for Disease Control (CDC).   These reports provide a detailed breakdown of US deaths by method and intent.  Using these reports, I was able to compile 30 years of data regarding firearm-related deaths.

There were two interesting findings derived from this effort.  First, the homicide rate in the US has reduced by 50% over the past 20 years.  While recent events might create the impression that gun-related violence is on the rise, overall firearm violence continues to trend lower.  Second, a close relationship has existed between homicide rates and firearm-related deaths over the past 30 years.  While I haven’t demonstrated the relationship mathematically, visually the relationship is striking.  The only significant deviation appears to be due to the events of 9/11/2001.  This does not prove a causal relationship between firearms and homicides but it definitely indicates that further exploration of this relationship is warranted.I then compared these homicide rates with those of other countries.  For the comparison to be relevant, I sought out countries with similar levels of prosperity to the US.  Using the World Bank figures for gross domestic product per capita (or Purchasing Power Parity) I identified a set of sixteen countries with a PPP between 125% and 70% of the USA.  I excluded countries such as Luxembourg, Qatar and Brunei, as their relatively small populations might make comparisons irrelevant.I obtained mortality data for over 100 countries from the World Health Organization’s Statistical Information System (WHOSIS).  To get the cause and intent, I had to employ the raw data provided by the WHO.  Unfortunately, details regarding the method and intent were only effectively captured in the tenth revision of the International Classification of Diseases (ICD-10).  Therefore, only the date ranges of 1999 to 2010 are useful.   Furthermore, data is not complete for all countries for this date range.   For this reason, I selected data collected in 2006 to compare countries.

Once again, the findings were striking.  Homicide rates in the US were more than three times that of Finland, the second highest on the list. The rate of firearm-related homicides in the US was greater than 8 times that of Canada, the second highest rate of firearm-related homicides amongst this grouping of countries.In a second pass, I sought to identify countries with similar homicide rates as the US.  I ordered the reporting countries by homicide rate for 2006 and selected eight countries preceding and seven countries succeeding the US in that list.

Of these countries, five stand out particularly for the same relatively high proportion of firearm-related homicides: Mexico, Costa Rica, Nicaragua, Thailand and Argentina.In subsequent postings I will explore the attributes of these five countries to understand what characteristics they share in common.  An aspiration of mine would be to produce a model that accurately represents these relationships, but given the small set of countries I am working with I am not confident in the viability of that.For me, this first effort provides a strong indication that even though gun related homicides in the US have been consistently trending lower,  the level of violence, and particularly firearm-related violence is out of proportion with the prosperity and maturity of our society.The data for this analysis can be found here.

Posted in Sandy Hook Project | Tagged , , | Comments Off

The Sandy Hook Project

Like many people, I was horrified and deeply saddened by the events of December 14, 2012.  The death of innocent children in this way is heartbreaking.  The thought that a single individual could quickly achieve such slaughter is deeply disturbing. This tragedy made me question the interpretation and justification of the second amendment.  Should the right to bear arms outweigh public safety? Does an individual’s right to self defense supersede collective rights to safety?  Moreover, are these rights so sacrosanct that we can’t even begin to discuss boundaries around what weapons are appropriate for the public?Because I was not the only person raising these questions, demands for tighter controls on guns began to surface in the public forum.  Many advocates for gun rights responded immediately to this outcry with  statistics, anecdotal evidence and some frankly questionable facts.As someone who works with data professionally and is passionate about the ethical use of data, I found the evidence being presented to be wanting.   Graphs represented specific points in time or involved very specific and sometimes unrelated data sets.  Data sources were frequently not identified, nor were there descriptions of the methodology employed.  I didn’t think that an accurate picture was being painted but I had no evidence of my own to refute it.I initiated the Sandy Hook Project in response to this need for information.  The purpose of this project is to produce rigorous and transparent analyses of data pertaining to gun-related violence.  My goal is to move beyond the rhetoric and emotion and produce (hopefully) objective insight  into the relationship between guns and violence in the US.  I realize that objectivity will be challenging, which is why I want to share the methods and the data openly so others can validate or refute my findings as well as contribute their own.

How the project will work (I hope)

I’ve put the project on GitHub. (https://github.com/john-m-spens/SandyHookProject).  While it’s not designed as a data repository, I think the ubiquitous nature of GitHub and the control enforced through the code repository model will support effective collaboration.I have already begun by creating a data set and analysis of global homicide statistics.  This work has produced new questions which I will address in subsequent efforts.  As I complete each analysis, I will load the data, site the source and describe the findings.  Eventually, I hope other collaborators will join the project, producing their own data sets and/or performing analyses with the existing data sets.  I’m hoping that people will share responsibilities of data collection for other collaborators as well as performing their own analyses.  Collaborators will start by posting a topic; the testing of a hypothesis, the answering of a question or the exploration of a particular data set that involves the relationships between firearms, violence, gun control laws, economic divisions, race, etc. In describing each topic, the collaborator is expected to describe their goal, the methodology the will apply and a description of the data sets required to support the work. The author of the topic will either employ existing data sets, generate new sets themselves or place requests to the community to assist in the collection of data.My ultimate hope is that the work done on this project will help people of the US make informed decisions regarding the control and ownership of guns.

Posted in Sandy Hook Project | Comments Off

The Value of Rework

I recently attended the Gartner BI Summit in Los Angeles.  At the conference Stephen Brobst, CTO of Teradata, led a great presentation on Enterprise Data Warehousing at Target.  During the presentation, Mr. Brobst talked about employing a phased approach to implementing data warehouses.  He stressed the importance of completing a logical data model for the data warehouse prior to implementation.  He warned that without such a model, 20% of the effort expended in each phase would be rework of the existing implementation due to the lack of that logical model.I was struck by this metric, but not because I thought it was surprising or excessive.  In fact, this sounds like an entirely reasonable percentage to me and I readily accept it.  I was more struck by the fact that this level of rework should be considered is a bad thing.  Does this mean that the project experienced 20% wasted effort each phase, and that if the team had completed analysis and definition of the logical data model, this number would be zero?Not necessarily.  If a team is planning to adopt any sort of phased implementation model, they should expect some level of rework.  And this is a good thing!  By choosing to deliver a data warehouse solution incrementally, the development team has, accidentally or intentionally, created an opportunity to generate feedback.   Once software is put into their users hands, the team will have a chance to determine if the requirements they expressed and the designs they produced are accurate.  invariably they discover that while some features meet the user’s expectations, others did not.  Some reports that were requested might turn out to be incomplete; requiring additional information to meet the business needs.  Other reports or analytics might prove to be redundant.  A new, automated alert or dashboard might eliminate the need for a certain report.  And regardless of the efforts made to verify requirements prior to or during implementation, these discoveries are often only made once the solution is in the users hands.Fortunately, though, because the development team chose to adopt a phased approach, they have an opportunity to remedy that.  These changes can (and should) be introduced into subsequent phases of the project.   Therefore, the team should assume some level of rework associated with each phase of the project.  And while this may limit their ability to deliver new functionality in subsequent phases, it also enables them to deliver a higher quality product to their user community.So let’s assume that some portion of the 20% was dedicated to work that resulted from user feedback, and not purely from incomplete analysis and design.  The second point I would like to consider is the cost to complete the logical data model.  If the team failed to fully complete the model initially, I must assume that some analysis and design work was left on the table.  Whether the team consciously decided to defer some work, or merely missed some detail, the effort they expended on the initial analysis and design was less than if they had completed the model.  Therefore, that rework effort most likely represents some level of deferred, rather than wasted, effort.  And in many cases, deferring work to the last responsible moment will do more to reduce waste rather than generate it.For example.  Let’s say that your development team is currently working on the topic of Products within an enterprise data warehouse for a manufacturing company.  The team’s objective is to represent the cost of manufacturing products.  The cost calculation may be highly complex, involving the current market price of raw materials at the time of manufacture.  Instead of performing the required analysis and modeling up front, the team could decide to initially employ a standard cost for raw materials, not only making the model simpler but also deferring efforts to acquire and transform market pricing.  By making this decision, they recognize they are deferring some analysis, design and implementation work that will need to be addressed later.  At the same time, this decision might have enabled the team to more rapidly deliver a solution that is satisfactory for most use cases.  The cost model can then be addressed at a later point, when it is a higher priority for the product owner.  Therefore, the rework incurred at that point is really not additional effort.  And by properly planning, and employing Agile testing and refactoring practices, the team can minimize the additional effort involved in implementing this feature later in the project.Many still equate the work ‘rework’ as a failure to properly plan and design up front.  But rework is not synonymous with waste.  Martin Fowler once told a team of developers, “We call it software because it is soft.  It is pliable and can be changed without tremendous cost”  There is some level of up front design that is required to avoid excessive rework and waste, but we if we work too hard to avoid rework, we can often generate more waste.

Posted in Practices | Tagged , , | Comments Off