In statistics, missing data, or missing values, occur when no data value is stored for the variable in the current observation. Missing data are a common occurrence and can have a significant effect on the conclusions that can be drawn from the data.
Types of missing data
Missing data can occur because of nonresponse: no information is provided for several items or no information is provided for a whole unit. Some items are more sensitive for nonresponse than others, for example items about private subjects such as income.
Dropout is a type of missingness that occurs mostly when studying development over time. In this type of study the measurement is repeated after a certain period of time. Missingness occurs when participants drop out before the test ends and one or more measurements are missing.
Sometimes missing values are caused by the researchers themselves. If data collection was not done properly or if mistakes were made with the data entry (Ader, H.J., Mellenbergh, G.J. 2008).
And a great deal of missing data arise in cross-national research in economics, sociology, and political science because governments choose not to, or fail to, report critical statistics for one or more years (Messner 1992).
It is important to question why the data is missing, this can help with finding a solution to the problem. If the values are missing at random there is still information about each variable in each unit but if the values are missing systematically the problem is more severe because the sample cannot be representative of the population. For example: a research is done about the relation between IQ and income. If participants with an over average IQ do not answer the question ‘What is your salary?’ the results of the research may show that there is no association between IQ and salary, while in fact there is a relationship. Because of these problems, methodologists routinely advise researchers to design research so as to minimize the incidence of missing values (Ader, H.J., Mellenbergh, G.J. 2008).
Techniques of dealing with missing data
Missing data reduce the representativeness of the sample and can therefore distort inferences about the population. If it is possible try to think about how to prevent data from missingness before the actual data gathering takes place. For example in computer questionnaires it is often not possible to skip a question. A question has to be answered, otherwise one cannot continue to the next. So missing values due to the participant are eliminated by this type of questionnaire, though this method may not be permitted by an ethics board overseeing the research. And in survey research, it is common to make multiple efforts to contact each individual in the sample, often sending letters to attempt to persuade those who have decided not to participate to change their minds (Stoop et al. 2010: 161-187). However, such techniques can either help or hurt in terms of reducing the negative inferential effects of missing data, because the kind of people who are willing to be persuaded to participate after initially refusing or not being home are likely to be significantly different from the kinds of people who will still refuse or remain unreachable after additional effort (Stoop et al. 2010: 188-198).
In situations where missing data are likely to occur, the researcher is often advised to plan to use methods of data analysis methods that are robust to missingness. An analysis is robust when we are confident that mild to moderate violations of the technique's key assumptions will produce little or no bias, or distortion in the conclusions drawn about the population.
Imputation
If it is known that the data analysis technique which is to be used isn't content robust, it is good to consider imputing the missing data. This can be done in several ways. Recommended is to use multiple imputations. Rubin[citation needed] argued that even with a small number, m, of repeated imputations (m being equal or smaller than 5) the quality of estimation improves enormously (in: Ader, H.J., Mellenbergh, G.J. 2008). For most practical purposes 2 or 3 imputations capture most of the relative efficiency that could be captured with a larger number of imputations. However, low values of m can lead to a substantial loss of statistical power, and some scholars now recommend that m be set to values from 20 to 100 or more (Graham, Olchowski, and Gilreath 2007). Obviously, any multiply imputed data analysis has to be repeated for each of the m imputed data sets and, in some cases, the relevant statistics have to be combined in a relatively complicated way (Ader, H.J., Mellenbergh, G.J. 2008). Examples of imputations are:
Partial imputation
The expectation-maximization algorithm is an approach in which values of the statistics which would be computed if a complete dataset were available are estimated (imputed), taking into account the pattern of missing data. In this approach, values for individual missing data-items are not usually imputed.
Partial deletion
Methods which involve reducing the data available to a dataset having no missing values include:
- Listwise deletion/casewise deletion (albeit a naive solution)
- Pairwise deletion(albeit a naive solution)
Full analysis
Methods which take full account of all information available, without the distortion resulting from using imputed values as if they were actually observed:
Interpolation
Main article:
InterpolationIn the mathematical field of numerical analysis, interpolation is a method of constructing new data points within the range of a discrete set of known data points.
See also
References
- Adèr, H.J.(2008). "Chapter 13: Missing data". In Adèr, H.J., & Mellenbergh, G.J. (Eds.) (with contributions by Hand, D.J.), Advising on Research Methods: A consultant's companion (pp. 305-332). Huizen, The Netherlands: Johannes van Kessel Publishing. ISBN 90-79418-01-3
- Graham J.W., Olchowski A.E., Gilreath T.D. (2007). "How Many Imputations Are Really Needed? Some Practical Clarifications of Multiple Imputation Theory". Preventative Science 8 (3): 208–213. doi:10.1007/s11121-007-0070-9.
- Messner SF (1992). "Exploring the Consequences of Erratic Data Reporting for Cross-National Research on Homicide". Journal of Quantitative Criminology 8 (2): 155–173.
- Stoop, I., Billiet, J., Koch, A., and Fitzgerald, R. (2010) Improving Survey Response: Lessons Learned from the European Social Survey. Wiley. ISBN 0-470-51669-0
- Zarate LE, Nogueira BM, Santos TRA, Song MAJ (2006). "Techniques for Missing Value Recovering in Imbalanced Databases: Application in a Marketing Database with Massive Missing Data". IEEE International Conference on Systems, Man and Cybernetics, 2006. SMC '06.. 3. pp. 2658–64. doi:10.1109/ICSMC.2006.385265. http://ieeexplore.ieee.org/xpls/abs_a ll.jsp?tp=&arnumber=4274271&i snumber=4274116.
Further reading
- Rubin, Donald B.; Little, Roderick J. A. (2002). Statistical analysis with missing data (2nd ed.). New York: Wiley. ISBN 0-471-18386-5.
- Enders, Craig K. (2010). Applied Missing Data Analysis (1st ed.). New York: Guildford Press. ISBN 978-1-60623-639-0.
- Allison, Paul D. (2001). Missing Data (1st ed.). Thousand Oaks: Sage Publications, Inc. ISBN 978-0-7619-1672-7.
- Acock AC (2005). "'Working With Missing Values". Journal of Marriage and Family 67 (4): 1012–28. doi:10.1111/j.1741-3737.2005.00191.x.
- Van den Broeck J, Cunningham SA, Eeckels R, Herbst K (October 2005). "Data cleaning: detecting, diagnosing, and editing data abnormalities". PLoS Med. 2 (10): e267. doi:10.1371/journal.pmed.0020267. PMC 1198040. PMID 16138788.
- Schafer, J. L.; Graham, J. W. (2002). "Missing data: Our view of the state of the art". Psychological Methods 7 (2): 147–177. doi:10.1037/1082-989X.7.2.147. PMID 12090408. – edit
- Graham, John W. (2009). "Missing Data Analysis: Making It Work in the Real World". Annual review of psychology 60: 549–576.
- Rubin DB (1976). "Inference and missing data". Biometrika 63 (3): 581–92. doi:10.1093/biomet/63.3.581.
External links
Background
Software
