Why is validity important and how can we improve it?

Validity refers to whether in an experiment a tool actually measures what it claims to and whether the results could actually be generalised to the wider world. There are two main types of validity ; internal and external.
Internal validity relates to whether changes made to a dependent variable actually change and influence the independent variable – in other words whether the tool measures what it is meant to. If there is high internal validity, you could expect the dependent variable to manipulate the independent variable, whereas if there was low internal validity it would be suspected that confounding variables were playing a part in effecting the independent variable.

There are various ways to assess to internal validity:
Face validity indicates whether a measure tests what it is meant to. Face validity can be low if for example in a questionnaire there is leading questions that influence a participant to answer a specific way.
Concurrent validity suggests whether a new test produces results that are similar to an existing test in the same field. If the test produces similar results to the existing valid tool than it is presumed to be valid.
Predictive validity
indicates whether a new measure can predict future consequences.

Internal validity can be improved in a few simple ways. In order to ensure an investigating is measuring what it is meant to, investigators can use single and double-blind techniques. A single-blind study is where the participant does not know the condition they are in, and double-blind is where neither the participant nor the experimenter knows what the groups represent. This method ensures there is little demand characteristics, such as trying to behave a certain way because they think that is what is expected of them, and also decreases experimenter effects, as they can not even accidently have a bias to a certain group. Another process that is effective when a repeated measure has been used is counterbalancing and random allocation, as they remove order effects, such as boredom, fatigue.

External validity on the other hand, refers to things that happen outside of the investigation and can effect whether findings are representative and can be generalised.
Historic validity questions whether if a measurement was used again in the future it would produce the same results. If the results are similar it means there is external high validity, as the tool is not susceptible to change.
Population validity can be tested by repeated an investigation on a different population/culture to decipher whether the findings can be generalised to different groups.
Ecological validity assess whether different settings and situations affect the findings.

To improve external validity, many investigators focus on the sample, and try taking larger, more varied samples, to try make the findings more applicable to the wider population. Additionally ecological validity is improved by carrying out the investigation in different, more realistic settings to try make the findings more relevant to everyday life.

Validity is essential because it is the basis of our conclusion – if there is low validity we cannot easily trust the conclusion of an investigation. An investigation that has no external validity should not be generalised to a wider population, as it may have just been the situation, timing or sample that produced the results.  Validity is important, especially internal validity, because without it our findings would not mean anything, as the investigator may not even be measuring what he sent out to, but instead a confounding variable.


16 Comments (+add yours?)

  1. screamingdominos
    Nov 24, 2011 @ 19:10:22

    You have outlined validity really well, not always easy, due to its variations, however I would like to contribute to your section on improving external validity. Good external validity means its representative of the target population, the best way to do this is by using random sampling over non-random samples, this means you are more likely to get participant that accurately represent the demographics of the target population which are usually varied. A common view is the more participants the more external validity although this is true the bigger the sample the better the generalizability bit random sampling can be used when a large sample is difficult to achieve.
    But over all the best way to improve external validity is to replicate the study over and over changing the who and the where, this can show that the results are representative of the target populations and the different situations associated with them.


  2. Trackback: Homework for my TA, Week 9 « screamingdominos
  3. psud5c
    Nov 25, 2011 @ 20:23:47

    This is a good in depth blog into the types of validity. An addition to your point about improving external validity you could have added how to avoid threats to external validity, which also would be a way to improve it. An example of a threat to external validity could be if a researcher is observing behaviour of a very specific group of participants for instance white children who attend a private school. The behaviours observed would most likely be very different to Black and Hispanic children attending a public city school. This limited range of the sample is a treat to external validity.


  4. Trackback: Homework for my TA Week 9 | psud5c
  5. kudzanai muchinhairi
    Dec 19, 2012 @ 16:58:44

    edit and correct on the first paragraph sentence that says “the dependent variable to manipulate the independent variable”. it is suppossed to be vice-versa.


    • Broadway2
      Oct 19, 2017 @ 00:27:32

      Exactly. I got confused the moment I got to that line…


      • Monyatsi Tomeletso
        Oct 19, 2017 @ 07:10:10

        very true. it is the manipulation of the independent variable (e.g. in farming fertilizer application rate across seasons) that affect the dependent variable (e.g. maize yield).

  6. Nyceer Omar
    Feb 04, 2014 @ 10:02:26

    Validity is a process that include gathering, interpreting, and using imformation


  7. hotels on
    Mar 02, 2014 @ 12:50:13

    Well done on acquiring essentially the most advanced blogs I have come across in many period! The just incredible how much you could make off from some thing due to precisely how visually wonderful it really is. Youve gather a fantastic blog room -great design, video, layout. This is certainly some sort of must-see blog!


  8. rekha begum
    May 13, 2015 @ 10:29:18



  9. rekha begum
    May 13, 2015 @ 10:29:52

    i think this is great but rekha is a fat cow


  10. rekha begum
    May 13, 2015 @ 10:31:35



  11. Monyatsi Tomeletso
    Mar 01, 2016 @ 05:17:28

    let me add to your description of PREDICTIVE VALIDITY.Predictive validity also refers to the degree to which a test forecasts some future behavior of a candidate, for instance academic performance in an accounting course. Predictive validity is computed by issuing an Intelligent Quotient (IQ) test and comparing the results with future scores in Standardized Achievement Tests (SAT). If IQ scores are lower than SAT scores, a candidate is over performing, if IQ scores are higher than SAT scores a candidate is under performing and if IQ scores equals SAT scores, then the candidate is performing as expected (Oosterhof, 2003; Kubiszyn & Borich, 2003)


  12. Broadway2
    Oct 19, 2017 @ 00:25:41

    Is it not the changes to the independent variable that affects dependent variables?
    …”Internal validity relates to whether changes made to a dependent variable actually change and influence the independent variable – in other words whether the tool measures what it is meant to.”


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: