Free CSSC Green Belt Exam Flashcards

Yield = ((opportunities - defects) / opportunities) x 100. Look the yield percentage up in a sigma table to convert it to a sigma level; a 99.5% yield sits between 4.0 and 4.1 sigma.

Sigma level is only a high-level snapshot. Comparing sigma levels across processes can point at the wrong target, so leadership must also weigh cost, resources, and the estimated impact of the improvement.

Customer-focused improvement, value streams, continuous process improvement, variation, removing waste, equipping people, and controlling the process.

Lack of support and buy-in at all levels, lack of resources or knowledge, poor project execution, data access issues, and doubt that Six Sigma fits a specific industry.

Carl Friedrich Gauss developed the normal curve in the 19th century. It is the most widely used probability model but only one of several, so never assume data is normal without testing it.

Motorola measured defects against a thousand opportunities until CEO Bob Galvin demanded a tenfold quality gain in five years in the mid-1980s. That goal pushed measurement to defects per million opportunities.

The levels are White, Yellow, Green, Black, and Master Black Belt. Green Belts do most of the data collection under a Black Belt's direction, keep other day-to-day job duties, and may lead smaller projects.

A three-by-three grid crossing three performance levels (performer, process, organization) with three dimensions (management, design, goals). It locates where a performance problem actually lives before a fix is chosen.

JumpStart finds problems and solutions in a single fast session with no rigorous verification or statistical analysis. Keep it to low-risk internal changes; avoid it for enterprise, customer-facing, or compliance-bound processes.

Overproduction, correction, inventory, motion, conveyance, over-processing, and waiting. CSSC adds three further forms of waste: talent, ideas, and capital or cash.

Type I is non-value-added work that is currently essential, such as inspection required because the process makes defects; make it efficient until the cause is fixed. Type II is non-essential and can be removed immediately.

Sort, Straighten, Shine, Standardize, and Sustain. Sustain depends on everyone committing to the standard; without that commitment the workplace falls into a cycle of cleaning up after each lapse.

Each step produces only what the next step or customer needs, when it is needed. If the second machine can process one part per minute, the first is set to one part per minute, attacking inventory and overproduction waste.

DPMO = (defects in the sample / opportunities for a defect in the sample) x 1,000,000. Two errors across 90 forms of 10 fields each gives (2 / 900) x 1,000,000 = 2,222 DPMO.

FTY = good units produced / units entering the process. RTY = (units entering - (scrap + rework)) / units entering. RTY comes out lower because it exposes rework that FTY quietly counts as success.

The problem or output y occurs because some inputs or causes x are occurring, usually more than one. Data analysis or a 5 Whys session validates the y = f(x) relationship so the team fixes the causes rather than the symptom.

Where and when the problem occurred, a measurement of its magnitude tied to cost where possible, a plain-language description, and the metric used to measure it. It names no cause and proposes no solution.

The steps, processing time, interdependencies, and resources and assignment. Working all four layers stops a team from treating a one-line description as a real process definition.

Inputs, outputs, events, tasks, and decisions. All the components are related, so changing an input or a decision rule changes the output.

Suppliers, Inputs, Process, Outputs, Customers. It is a Define-phase map that a team can draft in one brainstorming session, and it is scalable from a very small process to an entire business.

Monitor process performance with metrics, understand how the process serves business goals, keep standard operating procedures current, and make sure operators have training and resources. In Six Sigma they also keep the control plan in place.

Information is inherent in any process, whether or not it is being captured. If a process is not measured, the team must build a way to collect the data rather than conclude that none exists.

It starts from a specific critical customer need, breaks that need into drivers, and converts each driver into a measurable quality requirement. That is how Voice of the Customer statements become metrics.

Internal failures, caught before delivery, and external failures, which surface after delivery and hit customer satisfaction directly. Both are costs of nonconformity: outputs that miss critical-to-quality requirements.

A 15-criteria scoring model, starting with sponsorship, used to rank candidate projects. Its weighting lets a team make some criteria count more than others instead of scoring every criterion equally.

A timeline technique that maps the required activities and their dependencies so a team can see which chain of work sets the schedule. It needs team input, so it is often built during Define rather than before the project.

A senior leader who owns the final result, coaches the team at the charter stage, secures funds and resources, and absorbs corporate politics. Even the Black Belt reports to the sponsor or champion.

DMAIC (Define, Measure, Analyze, Improve, Control) improves a process that already exists. DMADV (Define, Measure, Analyze, Design, Verify) develops a new process or product.

Phases give a project control and organization, keep everyone on task, and break an overwhelming problem into tolerable pieces. CSSC calls this eating the elephant one bite at a time.

A concise problem statement, critical-to-quality metrics, the names and roles of team members, internal and external customers, the sponsor or champion, and a project duration. It is an output of the whole Define phase.

A quick scoping tool that answers where, what, and when the problem is and, just as explicitly, is not. Writing the boundary down fixes scope and feeds the problem statement before Measure begins.

Discrete data is categorical: nominal, ordinal, or binary. Continuous data is measured in units and is more precise. Continuous data can be converted to discrete, but discrete usually cannot become continuous, so collect continuous where possible.

Repeatability is one appraiser getting the same reading on the same item with the same system. Reproducibility is different appraisers reaching the same reading. Gage R&R tests both before a team trusts its data.

Four main connectors: People, Process, Materials, and Procedure. Some Black Belts add Equipment and Environment. The summarized problem goes at the fish head.

A matrix for proposing and ranking solutions against the root causes verified in Analyze. Leaders must still weigh timelines, resources, and scope, and selected solutions usually need sponsor or leadership approval.

A pilot is a limited trial in a live environment. It caps wasted resources if the solution fails, confirms the expected results occur, keeps troubleshooting small, and collects feedback from staff outside the team.

A controlled process has little variation; a capable process also centers on the customer requirement. Sigma level is the number of standard deviations from the process center to the nearest specification limit, and Cpk = sigma level / 3, so Cpk 1.33 equals sigma level 4.

It shows whether two factors move together: no correlation, weak correlation, or positive or negative correlation. It never proves causation, because two variables can be closely related with neither causing the other.

A normality test that compares the observed frequencies in the data against the frequencies a normal distribution would predict. It replaces eyeballing a histogram, which misleads when the shape is ambiguous.

R runs from -1 to 1. The closer it sits to either end, the stronger the relationship; at 1 or -1 every point falls on the trend line, and 0 means no relationship at all. Pearson's formula is the usual calculation.

Binomial covers discrete data with exactly two independent outcomes per trial, such as pass or fail. Poisson covers counts spread randomly across time, distance, or another unit; the word 'per' in a metric hints at Poisson.

Lognormal is continuous, bounded at zero, and positively skewed, so it fits durations such as machine downtime or income spread. Weibull is a family of continuous distributions whose shape setting lets it imitate several other curves.

Reject the null hypothesis when the p-value is less than alpha. A p-value at or above alpha means you fail to reject the null, which is not the same as proving the null true.

Alpha measures the risk of a Type I error, rejecting a null hypothesis that was actually true. Beta measures the risk of a Type II error, failing to reject a null hypothesis that was actually false.

Power = 1 - beta. Setting beta at 0.10 or 0.20 gives power of 0.90 or 0.80. The higher power is less likely to produce a Type II error but usually demands a much larger sample.

Alpha, beta or power, and delta, the smallest difference worth detecting, often expressed in standard deviations. Even a correct sample size draws the wrong inference if the sample was not random.

All three take variable data. Use X-bar & R for subgroups smaller than 8, X-bar & S for subgroups larger than 8 where sigma is easy to calculate, and I-MR when the data cannot be grouped into reasonable subgroups.

All four take discrete data. p and np track nonconforming units, u and c track counts of defects. np and c require a constant sample size; p and u do not.

Understand the target audience, tell a story with text and images, and stay clear and concise. Include statistics only where they support the decision being asked for.

An inadvertent analysis error, or forcing an analysis to fit a desired outcome or conclusion. Catch the first by having another Six Sigma expert, often a Master Black Belt, review your analysis before you present; never allow the second, whatever the presentation needs.