Statistics Lies Assessing the value of Collected Data

William Menna's image for:
"Statistics Lies Assessing the value of Collected Data"
Image by: 

The use of faulty statistics and the employment of misleading data-gathering methods have tarnished the name of what can be an incredibly valuable field. One of the most common methods for misusing data or creating misleading data deals with flawed question construction. Whether intentional or unintentional, problems with the questions used to create data can have a huge influence on the final product.

Unclear questions can create confusion or misunderstanding in the respondent. Examples of unclear questions include the use of negative items, double-negatives, and conditional questions. Conditional questions use the form of, "If this...then that." Another issue deals with a lack of specificity in questions. For example, asking about events or actions that occur within a specific time frame under the condition, "in the last month," can be ambiguous. Some respondents may take that to mean "in the last 30 days," while others may take it to mean the last calendar month.

Additionally, the use of biased language can influence answers. The use of anything other than neutral language in questions can push respondents to give different answers, hurting the validity of the generated data. For example, when the name of the estate tax was changed to "death tax," the public opinion of it was changed while its concept and specifics remained the same.

Further, the range of possible responses to questions should be exhaustive and all-inclusive. An example of a question with responses that are not sufficient would be to ask for respondents for their favorite fruit, but listing only apple and orange as possible choices. An appropriate set of responses should also include banana, pineapple, peach, plum, pear, etc., and a category for "other." Limiting the responses to only two or three possibilities when many other answers are possible, the results are not necessarily accurate. Answers must be exhaustive and all-inclusive. If the possible responses do not meet this criterion, the data created from the question is at risk of being flawed and unreliable.

Finally, always be mindful of who funded and carried out a study. Polls conducted by Gallup, Zogby, Rasmussen, or other scientifically unbiased polling companies tend to be the most reliable. On the other hand, any poll that has been conducted by a political organization should be taken with a grain of salt as those who conducted the research were probably looking for a specific answer.

Always be mindful of the way that the statistics have been generated. They can be very valuable, but only if created properly.

More about this author: William Menna

From Around the Web