Psychological testing involves the quantification and measurement of some behavior in order to obtain a score that will predict some future behavior from the individual. Psychological tests can take the form of intelligence tests or non-intelligence behavioral tests and assessments. They're given by counselors, therapists, and other test users who have an interest in helping the test taker.
Research and Define Purpose
Define the purpose of the test, the test users and the test takers. The purpose of the test should include the reason why the test is being made (e.g. measuring suicidal tendencies as an example). The test users are who will be giving the test, such as Licensed Professional Counselor or Psychologists. The test takers are the group of people to whom the test will be administered, such as those with autism or those with severe depression.
Research previous scientists who have produced similar psychological tests or have published professional articles on the subject of your test. These can be found in paid subscriptions to psychology databases like PsycInfo or the Mental Measurements Yearbook. Free resources would include access that libraries or universities give to members of various psychological journals.
Define the constructs of your psychological test. This involves defining what areas of intelligence or behavior you are testing, and a few examples are reading comprehension, processing speed, feelings of self worth, and suicide ideation. This should also involve researching how other scientists defined the construct you're attempting to measure.
Writing the Test
Plan the test administration and scoring. Write out who will administer the test, how long the test taker will have to complete the test, and the scoring procedures. This involves figuring out the weight of each test question or section and and how many points will be assigned to each. Also, decide whether the test will be given a cumulative score or other type of score.
Plan how many sections on the test there will be and what type of questions will be on the test. Some examples of the types of test questions include multiple choice, true and false, matching, coding symbols to letters, or even essays.
Write the test questions. Ideally, the test should have multiple types of questions so that it will fully measure the construct. For example, for a processing speed psychological test there could be coding sections and symbol matching sections.
Administer your psychological test to a sample of the population that you're studying. This should be done so that you can develop norms and standardization scores for future test administration.
Obtain a measurement of Test-retest Reliability by re-testing the original sample that you used to develop the norms and standardizations. Depending on the construct you're measuring, your test-retest interval will vary (example: Test-retest interval for measuring suicidal tendencies could be 1 week to several weeks).
Obtain other measurements of reliability by using specific formulas and other common reliability measures. For example, Cronbach's Alpha is a widely accepted measure of reliability. Others include Split-Halves Reliability (comparing the two halves of your test) and Parallel Forms reliability (comparing your test with another established psychological test).
For help with writing test questions, enlist the help of psychology graduate students.