Comments on: Three Kinds of Measurement and Two Ways to Use Them http://www.developsense.com/blog/2009/07/three-kinds-of-measurement-and-two-ways/ DevelopSense Blog Tue, 10 Jan 2012 14:44:17 +0000 hourly 1 http://wordpress.org/?v=3.3.1 By: Joe Harter http://www.developsense.com/blog/2009/07/three-kinds-of-measurement-and-two-ways/comment-page-1/#comment-274 Joe Harter Mon, 27 Jul 2009 17:10:19 +0000 http://developsense.com/wordpress/?p=157#comment-274 Thank you for the links to both articles, Michael. They were both personally poignant as I've been working on different ways to talk about metrics lately.<br /><br />I, too, differentiate between "metrics" and "measurements". I renamed "measurements" to "statistics". Perhaps that is my sports enthusiasm, but I find sports to be a good analogy here.<br /><br />In North American football there is a metric called <a href="http://en.wikipedia.org/wiki/Passer_rating" rel="nofollow">Passer Rating</a>. This is based on a formula that takes into account touchdowns, completion percentage, total yards and interceptions. Then a number between 0 and 158.3 is assigned as the rating for a passer. <br />What is the benefit of this metric? I have no idea, really, but it seems that the media, fantasy football players, and Las Vegas are more interested in it than the players and coaches are. It seems to me that the numbers are not the best way of determining how good a passer is. Through my reading it appears that the NFL teams determine who their passer will be based on first-order measurement. They actually watch the passer in action and decide how well they will fit in their own team, and then let results (usually wins vs. losses) speak for themselves.<br /><br />Afterall, the statistics for a passer can be influenced by temperature, wind, and the skills of the other players on the field in addition to the passer's skills. I would rather rely on first-order measurement and ignore the statistics or metrics on my test team. I would rather manage by results:<br /><br />- Increased knowledge of the system.<br />- Ability of the project team to make smart go-live decisions.<br />- Well-written bug reports.<br />- Finding important bugs faster.<br /><br />One last point - the teams in the NFL invest a lot of money in statisticians, so the stats must be helpful, right? It appears, according to one source, that the reason for the statistics is to <a href="http://vault.sportsillustrated.cnn.com/vault/article/magazine/MAG1085591/1/index.htm" rel="nofollow">market the players</a>. That is the way I’ve seen number used in software testing as well. It is used to make it <b>look</b> like a group/tester is doing a good/bad job.<br /><br />Thanks again for the great article, Michael. Thank you for the links to both articles, Michael. They were both personally poignant as I've been working on different ways to talk about metrics lately.

I, too, differentiate between "metrics" and "measurements". I renamed "measurements" to "statistics". Perhaps that is my sports enthusiasm, but I find sports to be a good analogy here.

In North American football there is a metric called Passer Rating. This is based on a formula that takes into account touchdowns, completion percentage, total yards and interceptions. Then a number between 0 and 158.3 is assigned as the rating for a passer.
What is the benefit of this metric? I have no idea, really, but it seems that the media, fantasy football players, and Las Vegas are more interested in it than the players and coaches are. It seems to me that the numbers are not the best way of determining how good a passer is. Through my reading it appears that the NFL teams determine who their passer will be based on first-order measurement. They actually watch the passer in action and decide how well they will fit in their own team, and then let results (usually wins vs. losses) speak for themselves.

Afterall, the statistics for a passer can be influenced by temperature, wind, and the skills of the other players on the field in addition to the passer's skills. I would rather rely on first-order measurement and ignore the statistics or metrics on my test team. I would rather manage by results:

- Increased knowledge of the system.
- Ability of the project team to make smart go-live decisions.
- Well-written bug reports.
- Finding important bugs faster.

One last point – the teams in the NFL invest a lot of money in statisticians, so the stats must be helpful, right? It appears, according to one source, that the reason for the statistics is to market the players. That is the way I’ve seen number used in software testing as well. It is used to make it look like a group/tester is doing a good/bad job.

Thanks again for the great article, Michael.

]]> By: Thomas V http://www.developsense.com/blog/2009/07/three-kinds-of-measurement-and-two-ways/comment-page-1/#comment-273 Thomas V Wed, 22 Jul 2009 23:01:17 +0000 http://developsense.com/wordpress/?p=157#comment-273 The oft-stated insistence that project or career goals must be "SMART" (specific, measurable, attainable, realistic, and time-based) often results in a confusion of metric and measurement, I believe. <br /><br />I think you could also make an argument that "have a better haircut" is a perfectly reasonable measurement from a goal-setting perspective, without requiring you to take a quantitative survey of your friends.<br /><br /><a href="http://www.vaniotis.com/thomas/blog/?p=24" rel="nofollow">On REAL goals and SMART goals</a> The oft-stated insistence that project or career goals must be "SMART" (specific, measurable, attainable, realistic, and time-based) often results in a confusion of metric and measurement, I believe.

I think you could also make an argument that "have a better haircut" is a perfectly reasonable measurement from a goal-setting perspective, without requiring you to take a quantitative survey of your friends.

On REAL goals and SMART goals

]]>