¶ … Statistics The plant opening is only a few months away and the Board of Directors for ABC Complete Kitchens, Inc. is interested in learning more about what you recommend for plant productivity analysis. Specifically, the board members want you to identify and describe the tools and techniques that are available that will help the plant's executive team better conduct statistical analyses for plant productivity evaluation. Be sure to define the management information systems in-place that will provide the productivity data required.

What quantitative research techniques/statistics would you recommend for ABC Complete Kitchens, Inc.

What information would you expect the statistics to provide for ABC Complete Kitchens, Inc.

Analyze research quantitative techniques, business statistics, and data evaluation techniques.

ABC Complete Kitchens, Inc. will certainly be helped by using quantitative statistics to help it analyze its plant productivity. Research has consistently found that the most successful business are inevitably those that use statistical research to evaluate possible success of their products, to predict the future of their market place (which includes both their client's desires and external and internal environmental conditions), and the state of their competition. Statistical tests are also conducted for probability -- to see whether one product is more likely to be attractive, cheaper and so forth than another, as well as to test causality / correlation -- for instance to test best ways of motivating or implementing KPIs.

All statistical tests that the business can use with examples will be elaborated on in this essay.

Introductory Terminology

Random sampling

Random sampling is not always used. Sometimes when this is impossible, less scientific strategies such as using a convenience sample (a ready-at-hand population) are employed. But the most scientific technique is to randomly select a probabilistic sampling of people that exemplify one's desired population and conduct the study on them.

ABC may for instance want to investigate whether their clients will be interested in their specific brand of kitchen appliances that they intend to launch the coming year. They can open their staff book of names and choose every 6th name for instance, or conduct lots on the names of their employees. Either method (or many more similar to this) will yield a probabilistic sample of names of individuals whom they can use for their study.

Other terminology:

Mean, Mode, Normal sampling / distribution, Outlier

The mean = the average number of the population who display a certain result / indication

The mode -- the most often. For instance, ABC Kitchens discovers that results show that Monday is the day on which most workers doze off.

Normal Distribution = this is the shape of most studies where you have sufficient number of people to conduct the study on and the people are well-matched. Most statistical tests are made for this. You have some so-called paramedic ones that are specifically designed for 'skewed' populations

Outlier - individuals (or variables) that fit outside the norm

Fundamental Statistical Tests

T-Test

The t-test is a comparison statistical strategy that is conducted between at least 2 groups. It is for the most simple of experiments.

ABC Kitchens may use it in the following way:

Scenario

ABC wishes to test two different techniques that are both said to improve employee motivation: Kotter's 8 steps for change and Kaizen, otherwise known as TQM.

ABC Complete Kitchens can proceed in the following way. It randomly distributes its employees into two groups. One group will be tested with the Kotter technique, with the instructors fully trained beforehand and all conditions closely assessed to make sure that there is reliability and validity in the experiment. It is best too (but not necessary) that there be a double blind in that people who are conducting the experiment are not aware which group is receiving experiment and which is the control group. The experimental group receives the experiment; the control group receives no program whatsoever.

ABC can conduct its experiment in three different ways: they can either hold a control a and...

This is followed by a month of inconspicuous monitoring and evaluation of employees work whilst employees of both groups fulfill their regular routine. Investigation is conducted seeing whether TQM has achieved any spike in improvement in motivation. After a month there can be a cross-over where the control group receives the training and researchers assess whether any improvement has been achieved. The researchers can then continue to testing Kotter's method in this same way.
An alternate way of conducting the experiment can be by researchers testing TQM on one group and Kotter on the other and seeing whether any improvement has been gained.

All studies have to have the elements of validity and reliability in order to be scientifically credible.

Reliability:

The people must be closely matched and the procedure of the study must closely fit the objective. Study must be operationalized (i.e. all requirements clearly defined) and it must be congruent to its case. In this way, we know that the results are not an off-chance that is common to this situation alone and likely due to some error in study. Rather we are surer that the results can extend to other similar situations too. When we know, for instance, that TQM works in similar situations too, ABC's study is said to be reliable

Validity

This is when possible bias is excluded from study as much as possible. One makes sure, for instance, neither that the workers in the two groups are closely matched in terms of qualities; that the environment is conducive to not distracting workers; that the interviewers are not biased nor that they harass the workers; that the study is not too long in order that the workers do not get fatigued and so forth.

Different types of T-tests

ABC can conduct the one sample independent t-test where they have this one group that first one method is tested on (e.g. TQM). The group is then monitored to check for improvement. After a while, Kotter's technique is taught to the group and the group is monitored again.

Alternately, ABC can use dependent groups which are two or more with each group being taught a different method (or one acting as control).

ANOVA, MANOVA and MANCOVA

When the study is more complex or when ABC Kitchens -- which is a global enterprise- wants to test an innovative method on productivity in all of its localities, it would then select key people from each and conduct an ANOVA, MANOIVA, or MANCOVA

ANOVA

All experiments have at least one dependent and independent variable. The dependent variable is that which is effected / or contingent on the tool being used. The independent variable is the stratagem / method / tool that is implementing change. For instance, ABC wants to test whether a scorecard will increase worker incentive. The scorecard is the independent variable; the worker's incentive is the dependent variable.

ANOVA is used when we have more than two groups in the study: it has only one dependent variable (raise in incentive) whilst the scorecard may be tested on 30 different groups.

A MANOVA and MANCOVA have multiple dependent variables. In order to affirm that the scorecard does increase incentive they may also want to test for other conditions in order to rule out that outcome is not a flux. They may therefore test too for fatigue, hunger, boredom, absence from work and so forth.

As opposed to the MANOVA, the MANCOVA includes more than one independent variable (e.eg a score card and another technique may be compared to assess results)

The above were comparison experiments.

Sometimes, however, ABC Kitchens may want to test correlation i.e. To see whether adopting a ceritan method or action will induce the wanted result.

Scenario

ABC wants to cut down on expense. It has recently adopted lean management and decides to transfer to a vertical instead of horizontal hierarchy that it thinks may be better for its organization.

To test this, they first need to define wanted results and specify these wanted results. Their instrument (their tool for accomplishing this) may be lean management. They will go about this by testing a certain amount of people and then, at the end of a certain amount of time, seeing whether there is any correlation between their activities and desired outcome. Strong correlations show that a shift towards vertical management may be more inspiring in terms of operational efficacy.

The statistical methods that they will use will be either a chi-square or a Pearson's depending on whether the test is parametric (i.e. 'noraml distruction) or non-parametric. The main result of a correlation is called the correlation coefficient (or "r"). It ranges from -1.0 to +1.0. The closer r is to +1 or -1, the more closely the two variables are related.

If r is close to 0, it means there is no relationship between the variables. If r is positive, it means that as one variable gets larger the other gets larger. If r is negative it means that as one gets larger, the other gets smaller (often called an "inverse" correlation). The correlation report shows the statistical significance of the outcome.

The Pearson correlation works best with linear relationships, namely as one variable gets…